Package forester.4.2.0

[jonsterling]

forester.4.2.0

A tool for tending mathematical forests

---

• Homepage: https://sr.ht/~jonsterling/forester/
• Source repo: git+https://git.sr.ht/~jonsterling/ocaml-forester
• Bug tracker: https://todo.sr.ht/~jonsterling/forester
• Release notes: http://www.jonmsterling.com/jms-00WK.xml

---

🐫 Pull-request generated by opam-publish v2.3.1

[jonsterling]

Damn, it looks like these issues are cropping up again: #24159 + #24738 + jonsterling/ocaml-forester@f1b8b6d

I really am at a loss...

[avsm]

Looks very green now, so merging! You may want to announce this on https://discuss.ocaml.org, where we have a Community category and an announce tag for this purpose.

[jonsterling]

Thanks for the encouragement! I've posted an announcement here: https://discuss.ocaml.org/t/ann-forester-4-2/15043.

Regarding the build issues I experienced in multiple forester releases (linked) when checking lower bounds, each time it was seemingly caused by some subtle behaviour of Menhir; running the reproduction commands via Docker as instructed in the build log takes at least an hour on my machine, and after that I could not figure out what is necessary to test whatever change I might make to resolve the problem without needing another hour — considering that many attempts might be necessary as I pinpoint the issue, it seemed to me that I was being asked to spend at least several hours... The edit-to-debug ratio involved in that process is too extreme for me to undertake it on a regular basis, as seems to be required, so I will in the future be removing my dependency on Menhir.

I understand the problem may ultimately stem from a complicated interaction between Menhir's versioning practices and my own, so it may not ultimately be any one entity's fault — but life is too short for me to find out, so I plan to just hand-write my parsers in the future.

[avsm]

You're also not obliged to use a range of Menhir versions. If you've only tested on a particular version for a given release of Forester, you can constrain the dependency here to that exact version with a =, and that's totally fine too.

[jonsterling]

That is a fair point, but I am trying to be a good citizen in the package ecosystem, and making Forester non-installable (without a fresh switch) for users who happen to have an only slightly different release of menhir installed seems to be a bit brutal for my users.